We examine various measures of COVID-19 infections, hospitalizations and deaths, with an emphasis on data for Kentucky. We find that: Data on the number of new reported cases of the disease obtained from convenience samples (as opposed to representative random samples) is an inaccurate measure of the spread of the disease in the State. Using CDC data and national studies, it appears that there were ten times the number of infections in March than reported for Kentucky at the time and by September the State is still capturing only one out of two people infected.

A better measure of new cases can be obtained from model-based estimates of new daily cases that adjusts for the number of people being tested, the demographic characteristics of who is being tested, hospitalization rates, death rates, data on mobility, as well as the known biases in the reported data.

Document Type

Research Paper

Publication Date


Working Paper Number

Working Paper 33